Considering Severity in Health Technology Assessment: Can We Do Better?

Problems with NICE's severity weights

This essay examines the implications, plausibility, and justification of the severity weighting that NICE (The National Institute for Health and Care Excellence) has endorsed for technology assessments in the U.K. It argues that the assignment by NICE of additional weights to health conditions which involve a large absolute or proportional shortfall of future expected QALYs (Quality-Adjusted Life Years) as compared to those who do not have these health conditions is not well supported and has troubling implications. The literature concerned with attitudes toward prioritizing severity has found a variety of notions of severity, and it is questionable to what extent those studies bear on whether to assign greater weights to health states involving large absolute or proportional shortfalls. In addition, the severity weighting is not well supported by either egalitarian or prioritarian political philosophy, because it is concerned only with the future and focuses only on health rather than well-being in general.

Charting public views on the meaning of illness severity

BACKGROUND

Illness severity is a central principle in multiple priority-setting frameworks, yet there is a paucity of research on public views regarding the meaning of illness severity. This study builds on the findings of a Q methodology study with members of the public that identified four general viewpoints on the meaning of illness severity. Here, we investigate the support for those viewpoints among the Norwegian population.

METHODS

Following piloting, the online survey was distributed to a broadly representative sample of the population (March to April 2023). The viewpoints from the earlier Q study were converted into vignettes: Lifespan, Subjective, Objective, and Functioning and Quality of Life (FQoL). The main task in the survey comprised ranking the vignettes and scoring them on a 0-10 visual analogue scale. We describe vignette alignment (from weak to strong) based on four categorisations (C1 to C4). C1 placed all respondents on their top scored vignette(s); C2 required a score of ≥7; C3 was designed to resolve ties; and C4 (which describes vignette membership) required a score of ≥7, a gap of two between vignettes scored ≥7, and did not allow ties.

RESULTS

The survey was completed by 1174 individuals; those who completed in ≤3.5 min were excluded. Of the final sample (n = 1094), 98.1% scored at least one vignette ≥7. In C1, 40.2% were aligned with Lifespan, 32.4% with FQoL, 28.9% with Objective, and 16.3% with Subjective. Using the C4 criteria, 55.4% did not have vignette membership, 13.6% had membership with Lifespan, 13.1% with Objective, 11.4% with FQoL, and 6.5% with Subjective.

CONCLUSIONS

Severity is an ambiguous term among members of the public. Decisionmakers ought to bear this plurality of meanings in mind, and perhaps reconsider whether using a term as multifaceted as 'severity' is helpful in formulating precise and transparent priority-setting criteria.

Scoring the Value Assessment Framework for China: A Factorial Survey

OBJECTIVES

This study aimed to develop the scoring functions for the recently developed value assessment framework (VAF) for China, which comprises 12 attributes.

METHODS

We implemented a factorial survey among Chinese healthcare stakeholders from July to September 2022. A total of 240 hypothetical drug value profiles described by the VAF were grouped into 60 blocks and randomly assigned to respondents. Each respondent was assigned with 1 block, each presented in 3 disease scenarios of different levels of severity. For each profile, respondents were asked to assess the drug's value on a scale from 0 (lowest) to 10 (highest) and make 1 of the 3 insurance recommendations: cover, to be negotiated for coverage, or reject. Linear and logistic mixed-effects models were used to develop scoring functions for aggregating the value attributes.

RESULTS

A total of 365 respondents participated in the survey. 3968 responses from 331 respondents were included in the analysis. Most of the included respondents were under 45 (n = 256, 77.3%), females (n = 208, 62.8%), living in urban areas (n = 296, 89.4%), and with a bachelor's degree or higher (n = 303, 91.5%). Health benefits and safety carried more weights than other attributes in the scoring functions across disease scenarios. The value and probability of entering negotiation or receiving insurance coverage for the attribute profiles for severe/critical disease were higher than for mild/moderate disease.

CONCLUSIONS

The scoring functions of the VAF can be used to assess the value of a drug and its probability of entering negotiation or receiving insurance coverage in China.

Risk preferences over health: Empirical estimates and implications for medical decision-making

Mainstream health economic theory implies that an expected gain in health-related quality of life (HRQoL) produces the same value for consumers, regardless of baseline health. Several strands of recent research call this implication into question. Generalized Risk-Adjusted Cost-Effectiveness (GRACE) demonstrates theoretically that baseline health status influences value, so long as consumers are not risk-neutral over health. Prior empirical literature casts doubt on risk-neutral expected utility-maximization in the health domain. We estimate utility over HRQoL in a nationally representative U.S. population and use our estimates to measure risk preferences over health. We find that individuals are risk-seeking at low levels of health, become risk-averse at health equal to 0.485 (measured on a 0-1 scale), and are most risk-averse at perfect health (coefficient of relative risk aversion = 4.51). We develop the resulting implications for medical decision making, cost-effectiveness analyses, and the proper theory of health-related decision making under uncertainty.

Systematic Review of the Relative Social Value of Child and Adult Health

OBJECTIVES

We aimed to synthesise knowledge on the relative social value of child and adult health.

METHODS

Quantitative and qualitative studies that evaluated the willingness of the public to prioritise treatments for children over adults were included. A search to September 2023 was undertaken. Completeness of reporting was assessed using a checklist derived from Johnston et al. Findings were tabulated by study type (matching/person trade-off, discrete choice experiment, willingness to pay, opinion survey or qualitative). Evidence in favour of children was considered in total, by length or quality of life, methodology and respondent characteristics.

RESULTS

Eighty-eight studies were included; willingness to pay (n = 9), matching/person trade-off (n = 12), discrete choice experiments (n = 29), opinion surveys (n = 22) and qualitative (n = 16), with one study simultaneously included as an opinion survey. From 88 studies, 81 results could be ascertained. Across all studies irrespective of method or other characteristics, 42 findings supported prioritising children, while 12 provided evidence favouring adults in preference to children. The remainder supported equal prioritisation or found diverse or unclear views. Of those studies considering prioritisation within the under 18 years of age group, nine findings favoured older children over younger children (including for life saving interventions), six favoured younger children and five found diverse views.

CONCLUSIONS

The balance of evidence suggests the general public favours prioritising children over adults, but this view was not found across all studies. There are research gaps in understanding the public's views on the value of health gains to very young children and the motivation behind the public's views on the value of child relative to adult health gains.

CLINICAL TRIAL REGISTRATION

The review is registered at PROSPERO number: CRD42021244593. There were two amendments to the protocol: (1) some additional search terms were added to the search strategy prior to screening to ensure coverage and (2) a more formal quality assessment was added to the process at the data extraction stage. This assessment had not been identified at the protocol writing stage.

On the (Near) Equivalence of Welfarist and Extra-Welfarist Methods to Value Healthcare With Implications for Assessing Equity

OBJECTIVES

While welfarist economics (WE) methods rely wholly on individuals' valuations, extra-welfarist (EW) methods seek alternative measures of value. Major reviews of the EW literature conclude that EW studies almost universally replace "utility" with "health" as the maximand. This analysis seeks to understand what conclusions are necessary and sufficient to make EW and WE methods concurrent and discusses implications for measuring social value.

METHODS

Using standard WE methods, I demonstrate that EW is equivalent to WE with 2 key restrictions-individuals have constant returns to health in producing utility and health budgets are fixed. Fixing budgets removes a key WE step, determining the marginal rate of substitution between consumption and health, the willingness to pay for health gains.

RESULTS

Because EW methods equate with WE with these 2 restrictions, I show how formal models to construct aggregated social welfare functions (SWFs) in WE frameworks lead directly to SWF models using EW models of value. I also show that, in fixed-budget health systems, when SWFs place different values for improving health of different subpopulations, aggregate health output fails as a SWF criterion. I demonstrate how different societal values can and should enter EW SWF models using WE criteria. I also discuss the implications when either of these key restrictions does not properly represent people's preferences.

CONCLUSIONS

Once EW methods are shown to be a restricted form of WE methods, those WE methods can illuminate how best to measure SWFs in EW environments.

Rule of Prevention: a potential framework to evaluate preventive interventions for rare diseases

Background: The benefits of preventive interventions lack comprehensive evaluation in standard health technology assessments (HTA), particularly for rare and transmissible diseases. Objective: To identify possible considerations for future HTA using analogies between the treatment and prevention of rare diseases. Study design: An Expert panel meeting assessed whether one HTA assessment framework can be applied to assess both rare disease treatments and preventive interventions. Experts also evaluated the range of value elements currently included in HTAs and their applicability to rare, transmissible, and/or preventable diseases. Results: A broad range of value should be considered when assessing rare, transmissible disease prevention. Although standard HTA can be applied to transmissible diseases, the risk of local outbreaks and the need for large-scale prevention programs suggest a modified assessment framework, capable of incorporating prevention-specific value elements in HTAs. A 'Rule of Prevention' framework was proposed to allow broader value considerations anchored to severity, equity, and prevention benefits in decision-making for preventive interventions for rare transmissible diseases. Conclusion: The proposed prevention framework introduces an explicit initial approach to consistently assess rare transmissible diseases, and to incorporate the broader value of preventive interventions compared with treatment.

Methods to Adjust Willingness-to-Pay Measures for Severity of Illness

OBJECTIVES

Both private sector organizations and governmental health agencies increasingly use illness severity measures to adjust willingness-to-pay thresholds. Three widely discussed methods-absolute shortfall (AS), proportional shortfall (PS), and fair innings (FI)-all use ad hoc adjustments to cost-effectiveness analysis methods and "stair-step" brackets to link illness severity with willingness-to-pay adjustments. We assess how these methods compare with microeconomic expected utility theory-based methods to value health gains.

METHODS

We describe standard cost-effectiveness analysis methods, the basis from which AS, PS, and FI make severity adjustments. We then develop how the Generalized Risk Adjusted Cost Effectiveness (GRACE) model assesses value for differing illness and disability severity. We compare AS, PS, and FI against value as defined by GRACE.

RESULTS

AS, PS, and FI have major and unresolved differences between them in how they value various medical interventions. Compared with GRACE, they fail to properly incorporate illness severity or disability. They conflate gains in health-related quality of life and life expectancy incorrectly and confuse the magnitude of treatment gains with value per quality-adjusted life-year. Stair-step methods also introduce important ethical concerns.

CONCLUSIONS

AS, PS, and FI disagree with each other in major ways, demonstrating that at most, one correctly describes patients' preferences. GRACE offers a coherent alternative, based on neoclassical expected utility microeconomic theory, and can be readily implemented in future analyses. Other approaches that depend on ad hoc ethical statements have yet to be justified using sound axiomatic approaches.

A severely fragmented concept: Uncovering citizens' subjective accounts of severity of illness

Universal healthcare is constrained by national governments' finite health resources. This gives rise to complex priority-setting dilemmas. In several universal healthcare systems, the notion of severity (Norwegian: 'alvorlighet') is a key consideration in priority setting, such that treatments for 'severe' illness may be prioritised even when evidence suggests it would not be as cost-effective as treatment options for other conditions. However, severity is a poorly-defined concept, and there is no consensus on what severity means in the context of healthcare provision - whether viewed from public, academic, or professional perspectives. Though several public preference-elicitation studies demonstrate that severity is considered relevant in healthcare resource distribution, there is a paucity of research on public perceptions on the actual meaning of severity. We conducted a Q-methodology study between February 2021 and March 2022 to investigate views on severity amongst general public participants in Norway. Group interviews (n = 59) were conducted to gather statements for the Q-sort ranking exercises (n = 34). Data were analysed using by-person factor analysis to identify patterns in the statement rankings. We present a rich picture of perspectives on the term 'severity', and identify four different, partly conflicting, views on severity in the Norwegian population, with few areas of consensus. We argue that policymakers ought to be made aware of these differing perspectives on severity, and that there is need for further research on the prevalence of these views and on how they are distributed within populations.

The new World Mental Health Report: Believing impossible things

OBJECTIVE

We examine whether the recent World Health Organization (WHO) report on global mental health uses severity of illness as a criterion in priority setting for resource allocation.

CONCLUSIONS

The WHO does not prioritise severity in the recent landmark World Mental Health Report. It recommends instead the insuperable task of scaling-up interventions for broadly defined mental health conditions, including milder distress, amongst over a billion people, with the majority living in low- and middle-income countries. Schizophrenia, the most severe and disabling of all psychiatric illnesses, is relatively neglected in the WHO report, and the disability associated with bipolar disorder is underestimated. This is inconsistent with the ethical principle of vertical equity, where the most severe illnesses should receive the greatest priority. The global mental health movement must refocus on deinstitutionalisation, and ensure adequate community and general hospital treatment for severe illnesses, especially the 24 million people with schizophrenia.

Spatial analysis and factors associated with transcatheter aortic valve implantation in Portugal: a retrospective analysis from 2015 to 2017

OBJECTIVES

To identify the factors associated with transcatheter aortic valve implantation (TAVI) use of TAVI in inpatients with aortic stenosis (AS) in Portugal and its geographical distribution.

METHODS

A quantitative, observational and retrospective study using the Portuguese National Health Service inpatient discharge database from 2015 to 2017. Surgical aortic valve replacement (SAVR) and TAVI procedures were selected using the International Classification of Diseases. First, we mapped the yearly age-standardised rate for each procedure using QGIS. Then, we performed χ² tests, independent t-tests and logistic regressions to study the factors associated with TAVI use.

RESULTS

From 2015 to 2017, 8398 hospitalisations were selected, 88.5% SAVR and 11.5% TAVI. From 2015 to 2017, SAVR use increased in the Northern region and decreased in the Lisbon region, while the opposite was observed for TAVI. TAVI was performed among the most complex (p<0.001) and older patients (the mean (SD) age for SAVR was 70 (±11) years old and 81 (±7) years old for TAVI, p<0.001). The results for the logistic regressions showed that, more recent hospitalisations, being older, living in the Lisbon region and having a higher Charlson Comorbidity Index was associated with an increased likelihood of undergoing TAVI (p<0.001).

CONCLUSIONS

TAVI increased over the years. TAVI is more often performed in more severe patients as an alternative to SAVR with similar discharge outcomes. These results suggest the existence of geographic disparities in the availability and access to healthcare services and technologies.